1. Field of the Invention
The present invention relates generally to an evaluation value computing system in a production line simulator. More particularly, the invention relates to a method for determining a parameter to be set in a simulator in a production line simulation.
2. Description of the Related Art
Conventionally, when a simulation of a production line is used, modeling of a current line is performed. In modeling of the production line, parameters to be set in a model are determined on the basis of experience and/or prediction of an operator responsible for setting the parameters. In the alternative, a method to determine the parameters on the basis of collected performance data has also been known.
Here, as parameters to be set in the simulator, a time required in process in each production step of the production line, a time requiring for preparation for the process, a time for constraining a production facility, tool, and worker, a time required for transporting a product to a next process step after completion of operation in each production step, a time required for moving the tool and/worker and so forth are known as important parameters. These times are generally referred to as xe2x80x9cstandard timexe2x80x9d.
Upon implementation of the production line simulation, a value representative of how long a product in question constrains a production facility in question when the product is processed in the production facility, namely the xe2x80x9cstandard timexe2x80x9d is important factor determining reliability of a result of simulation. Since the xe2x80x9cstandard timexe2x80x9d significantly influences for the result of simulation, the xe2x80x9cstandard time xe2x80x9d may also be referred to as xe2x80x9cevaluation valuexe2x80x9d.
The xe2x80x9cstandard timexe2x80x9d as the evaluation value fluctuates significantly depending upon process condition, amount of products and so forth. Therefore, it is determined or derived by the method set forth above on the basis of time-series information of past events occurred in the factory and set as one of parameter in the production line simulation. Such determination or derivation method utilizes one of or combination of an average value, a maximum value, a minimum value, a center value, a most frequent value and dispersion, or a value derived by the foregoing method and fluctuation assuming normal distribution.
In Japanese Unexamined Patent Publication No. Heisei 7-129677, a method of deriving the evaluation value from a production performance data by means a reference data generating means is employed. On the other hand, in Japanese Unexamined Patent Publication No. Heisei 7-45490, when a difference between the simulation result and performance is large, for a parameter having large dispersed value, a method of automatically re-deriving a value is employed. Furthermore, while not employed in the simulation, as a method of deriving the evaluation value reflecting a distribution, there is a derivation method employing histogram.
A first problem is that when. the evaluation value derived by the conventional method is set, difficulty is caused in matching the simulation result and the performance. The reason is that the evaluation value derived by the conventional method is not appropriate for representing its parameter, in practice. Namely, the reason is that the conventional method is established under a premise that the distribution of the performance data is symmetric or normally distributed. In practice, distribution of the data is irregular shape and contains large number of abnormal value. As a result, in order to obtain simulation result adapted to performance, adjustment of parameter with huge amount of process steps with human judgment is required.
Even in the method disclosed in Japanese Unexamined Patent Publication No. Heisei 7-45490, it cannot be adapted to the case where a plurality of evaluation values are not certain. Furthermore, in order to determine one evaluation value, simulation has to be performed for many times. In simulation of the production line, it has been know that the final simulation result should fluctuate significantly if the evaluation value of the primary one of production facility is not certain. This appears more significantly in a semiconductor fabrication line.
A second problem is that when an abnormal value is contained in the data as objective for analysis in the conventional method, the evaluation value should be affected significantly by the abnormal value unless the abnormal value is removed preliminarily. Therefore, a logic for eliminating the abnormal value has to be preliminarily prepared. On the other hand, there is no means for definitely determining a range of the abnormal value other than that setting by human experience.
A third problem is to take a long period until execution of reliable simulation utilizing a method reflecting the conventional distribution in order to derive the evaluation value. The reason is that, in the method of the conventional histogram reflecting distribution, there is no simple means for determining a unit width of the histogram and where to start the initial value, and it takes a long period for computing process to require huge amount of time for obtaining a final evaluation value.
A fourth problem is that, particularly in the semiconductor fabrication, it becomes necessary to review once derived evaluation value for frequent occurrence of fluctuation of mixing of product class, addition of new production facility, expansion or modification of condition. Then, set parameter has to be adjusted with huge amount of process step at every occasion of reviewing.
It is therefore an objection of the present invention to provide a simulation model by deriving an evaluation value of parameter adapted to a real data, an evaluation value computing system on a line simulator of a production line, which can obtain realistic simulation result, and an evaluation value computing method.
According to the first aspect of the present invention, an evaluation value computing system in a production line simulator for computing a standard time of a product processing period as an evaluation value, which is set in a production facility model to be used in the production line simulation and represents a period for constraining state of the production facility for production process or preparatory work for production process, comprises:
event reporting means for reporting event occurring in the production facility;
required time deriving and determining means for deriving and determining a required time required for production process in the production facility on the basis of a reported event, and accumulating the required time as data; and
evaluation value computing means for deriving the standard time on the basis of the accumulated data, the evaluation value computing means including:
sorting means for sorting the accumulated data in a sequential order of values,
center value means for deriving a first center value on the basis of result of sorting;
zone setting means for detecting data on which an absolute value of a difference with the first center value becomes smaller and setting a zone where number of the data to be detected exceeds half of all data number, and
evaluating means for deriving a new second center value of data within the set zone, comparing a difference between the second center value and the first center value with a convergence judgment reference value preliminarily set close to zero and determining whether the second center value can be regarded as the evaluation value on the basis of result of comparison.
In the preferred construction, the evaluation value computing means may repeatedly operate the zone setting means and the evaluation means until the difference becomes smaller than the convergence judgement reference value when the difference between the second center value and the first center value is greater than the convergence judgement reference value.
The evaluation value computing system may further comprise data classifying means for classifying data determined by the required time deriving and determining means into a computing unit of the evaluation value depending upon production facilities, product processing conditions and products. Also, the evaluation value computing system may further comprise extraction means for narrowing down objective data for deriving the evaluation value in time among accumulated data. Furthermore, the evaluation value computing system may further comprise evaluation value setting means for setting the derived evaluation value to the production line simulator.
According to the second aspect of the present invention, an evaluation value computing method in a production line simulator for computing a standard time of a product processing period as an evaluation value, which is set in a production facility model to be used in the production line simulation and represents a period for constraining state of the production facility for production process or preparatory work for production process, comprises:
event reporting step of reporting event occurring in the production facility;
required time deriving and determining step of deriving and determining a required time required for production process in the production facility on the basis of a reported event, and accumulating the required time as data; and
evaluation value computing step of deriving the standard time on the basis of the accumulated data, the evaluation value computing step including:
sorting step of sorting the accumulated data in a sequential order of values,
center value step of deriving a first center value on the basis of result of sorting;
zone setting step of detecting data on which an absolute value of a difference with the first center value becomes smaller and setting a zone where number of the data to be detected exceeds half of all data number, and
evaluating step of deriving a new second center value of data within the set zone, comparing a difference between the second center value and the first center value with a convergence judgment reference value preliminarily set close to zero and determining whether the second center value can be regarded as the evaluation value on the basis of result of comparison.
Preferably, in the evaluation value computing step, the zone setting step and the evaluation step may be repeated until the difference becomes smaller than the convergence judgement reference value when the difference between the second center value and the first center value is greater than the convergence judgement reference value.
The foregoing method may further comprise data classifying step of classifying data determined by the required time deriving and determining step into a computing unit of the evaluation value depending upon production facilities, product processing conditions and products. The method may further comprise extraction step of narrowing down objective data for deriving the evaluation value in time among accumulated data. Also, the method may further comprise evaluation value setting step of setting the derived evaluation value to the production line simulator.
According to the third aspect of the present invention, a storage medium storing a control program for an evaluation value computing method in a production line simulator for computing a standard time of a product processing period as an evaluation value, which is set in a production facility model to be used in the production line simulation and represents a period for constraining state of the production facility for production process or preparatory work for production process, the evaluation computing method comprises:
event reporting step of reporting event occurring in the production facility;
required time deriving and determining step of deriving and determining a required time required for production process in the production facility on the basis of a reported event, and accumulating the required time as data; and
evaluation value computing step of deriving the standard time on the basis of the accumulated data, the evaluation value computing step including:
sorting step of sorting the accumulated data in a sequential order of values,
center value step of deriving a first center value on the basis of result of sorting;
zone setting step of detecting data on which an absolute value of a difference with the first center value becomes smaller and setting a zone where number of the data to be detected exceeds half of all data number, and
evaluating step of deriving a new second center value of data within the set zone, comparing a difference between the second center value and the first center value with a convergence judgment reference value preliminarily set close to zero and determining whether the second center value can be regarded as the evaluation value on the basis of result of comparison.
Discussing in connection with the operation of the present invention set forth above, the required period necessary for performing production process in the production facility is derived and determined on the basis of event occurring timing, event receiving timing or state continuing period contained in the event from the event reported from each production facility of the production line. The required period thus determined is accumulated as data. Then, the xe2x80x9cstandard timexe2x80x9d as evaluation value is derived from the accumulated data. As a method for deriving the evaluation value, the accumulated data is sorted in sequential order of values. Then, from the result of sorting, the first center value is derived. Then, data on which an absolute value of a difference with the first center value becomes smaller is detected. A zone is then set so that number of the data to be detected exceeds half of all data number. The new second center value of the data within the set zone is derived. The difference between the second center value and the first center value is compared with a preliminarily set reference value which is set close to zero to determine whether the second center value can be regarded as the xe2x80x9cstandard timexe2x80x9d as the evaluation value. Thus, the evaluation value, namely the standard time better reflecting the distributing condition of the live data can be obtained to provide higher reliability for the result of the production line simulation.